In light of recent endeavours.. While this booster is currently requiring linking back to the STE 32MHz clock and new firmware.. Oddly a firmware I created does work at 36MHz.. But this requires the CPU CLK inverting, which means the 8MHz clock is out of phase with the motherboard so I am not happy with that solution.
But also, reprogramming a 22V10 GAL hundreds of times as a total nightmare as I keep having to remove it from the socket, but it in the programmer, but it back in the socket, with my poorly fingers I just had to stop doing that.
What I have started tonight is a ATF1502 PLCC 44 PLD to 22V10 adapter... Which will basically be the same as there Altera board I did ages ago..
viewtopic.php?f=24&t=76#p202 This way I can do "in circuit programming".
But more to the point, I think the problem is the clock low times are being violated during switching transitions... I'm too lazy to take scope captures of this at the moment, but basically when switching from 32MHz to 8MHz, it always starts the 8 MHz cycle on the low part. And it always starts on the low part and finishes it in switching into 32MHz. When using the STE's 32MHz clock, this timing is always identical it is always stable. However when using a external oscillator such as 36MHz, that low period is basically random and that is that I believe where things are going wrong.
The way to solve this is to have both clocks being used to switch the frequencies, however a limitation of the 22V10 is it can only have one clock input to drive the flip-flops, which has to be 32MHz, and there is no way to even manually create this delay as the 22V10 simply doesn't have enough LA's or IOs. So moving to more "mature" PLD such as the ATF1502 is the only way I can create proper sync logic.
I have done some tests by delaying the 8MHz clock by a "double inverter" and also use that in the sync logic, basically just adding a few ns (about 40ns) so it stretches the clock edges out of it, so when I sync to them, they are actually slightly out of sync to what the ST clock is doing. Basically a type of "pulse stretcher" so it is just enough to get the 36MHz running stable and the hard drive also works perfectly as well. But it is all just too much logic to fit into the 22V10 to do it all properly.
So future V1.5 batches will have to be linked to the STE's 32MHz clock like the V1 booster. I don't know when I will get chance to do more boards, I will be have to a third party to assemble a few.. Meanwhile I will finish designing this ATF adapter board, and in a couple of weeks time I can experiment with the code a lot better and see if I can bring back a 36MHz or 40MHz booster. Work on this will probably be somewhat slow though.
EDIT:
Crappy adapter PCB done.

- e671b9941312b1daf105d14bbac803e7.png (146.54 KiB) Viewed 2722 times